High temperature thermal energy Storage by Reversible thermochemical Reaction

Final Report Summary - STORRE (High temperature thermal energy Storage by Reversible thermochemical Reaction)

Executive Summary:
The StoRRe project concerns the area of thermal energy storage by chemical reaction for thermal solar power plants. The objective of the project is to develop a new and promising solution for the heat storage with the following characteristics:
− Mid-term (24h to few days). Long-term (several months) heat storage was originally considered and analyzed during early stages in the project but was considered unsuitable due to lower efficiencies.
− High storage density (300-500 kWh∙m-3),
− High temperatures (300-600°C), which are representative of the Concentrated Solar Power (CSP) plants with parabolic trough or Compact Linear Fresnel Reflectors (CLFR) technologies.
In the particular case of the StoRRe project, the investigation has focused on the reversible reaction: Ca(OH)2 solid + 100 kJ/mol ↔ CaOsolid + H2Ogas
The final results are not very different from the objectives of the project, proving a substantial level of achievement of the overall objective of the StoRRe project. The main subjects of R&D activities were covered with success:
− The storage process modelling, determination of mass and heat balances and effective performance in terms of energy yield and storage density
− Heat-exchanger reactor technology concept and pre-design at industrial scale in the range of 100 MW
− Lime kinetics in the range of conditions expected by the process
− Lime material improvement for the process: identification of requirements, development, characterization and evaluation at laboratory and pilot scale
− Experimental proof-of-concept of the heat-exchanger reactor and reactor modelling validation, identification of the main reaction limitations.
− Integration in the solar plant, up-scaling, costs evaluation and final techno-economic assessment.
The thermochemical storage concept proposed in the Storre project has been brought from TRL 1 at the beginning of the project to TRL 3-4 at the end of the project.

Project Context and Objectives:
The StoRRe project had as its main objective to develop and experimentally demonstrate at realistic conditions a novel concept for thermochemical energy storage based on the reversible hydration reaction of CaO and dehydration reactions of Ca(OH)2, that was at a very early stage of development when the project was initiated. The project has required R&D activities in a wide range of scales and with different approaches: from kinetic determinations and material development work at particle level, to pilot design and testing of reactors with electric thermal inputs up to 20 kW, to modelling and simulation work of large reactors and processes of the full energy storage system rated in the range of 100 MW. Therefore, the summary of the activities that follows is organised following the Work Packages structure of the project.
STORAGE PROCESS MODELLING (WP3)
A simplified scheme of the storage process was modelled using 2 similar methods, EXCEL sheet or DYMOLA software. The 2 models were quasi-static models, this level of modelling was consistent with the level of knowledge of the storage system as basic interrogations concerning heat and mass balances were questioned.
Several configurations, mainly daily or seasonal storage and pure steam or a mixture of air and steam as fluidising gas, were tested. Because of various energy losses in the system -steam production, solid heating, fluidising gas circulating power- the global energy yield of the storage process was under 50% in all cases, and even below 40% for some seasonal storage. Concerning the energy density of storage per cubic meter, taking into account the powder low density and the reaction conversion efficiency, the volume density of storage could be as low as 200 kWh/m3, under the Storre target.
These facts learned from the process modelling highlighted the need to consider short term storage (to a few days) so that both sensible and chemical heat could be charged and discharged in the process and the importance to better integrate the storage process in the solar plant, in particular to exploit better the steam produced during dehydration.
REACTOR MODELING (WP3)
The design of new thermochemical energy storage process requires a sufficient understanding of reactor performance, as it is necessary to know the output conversion of solids and gases from the reactor as a function of input streams, operating conditions and general dimensions and other design characteristics (internals etc.) of the reactor itself. Reactor models, once they are validated with experiments at suitable scale, become a valuable tool for design and scale up the process.
The Storre process will contain at least one fluidized bed reactor to undertake the hydration of CaO particles during the energy discharge mode of the system and the dehydration Ca(OH)2 loaded solids during the energy charge. The level of detail and complexity adopted to quantify the different phenomena inside a chemical reactor to undertake these reactions depends on the scale level of the reactors under testing. Within the objectives of the Storre project (demonstration level at TRL4), the “classic” bubbling bed reactor model of Kunii Levenspiel (1D) was adopted to interpret all experimental results from pilots and supply the design information for scaled up simulations of the process. The underlying assumptions in such model (perfect mixing of solids, two gas phases with a certain crossflow of gas to account for the exchange of gases between phases) have been shown to describe sufficiently well most of the experimental results obtained from batch and continuous experiments of WP6. The KL reactor model, adapted for the hydration and dehydration reactions by introducing kinetic sub-models adequate for the specific solids undergoing reaction in the bed, can explain reasonable well all the results using a single cross flow factor as adjustable parameter. The predictions of the reactor model have been used to estimate operating windows in reactors operating in larger scale processes developed in WP6 and WP7 and will be a valuable tool for further process development.
HEAT-EXCHANGER REACTOR PRE-DESIGN (WP3)
The ruling idea of the heat-exchanger concept at industrial scale is a simple geometry in which the solid handling and flow are as simple as possible. With the help of a simplified model written under Matlab and coupling the kinetics of the hydration or dehydration reaction established from the project (WP4) with the thermal transfers in the bed given by literature correlations, the general figures of the reactor were designed. If the kinetics is one of the key parameter, the tube bundle surface and the fluidising velocity at the base of the bed are also dimensioning the bed volume. The model highlighted the fact that, contrary to what was expected, the optimal size of lime particles in the process on a chemical and thermal point of view for a BFB was not very small particles but particles with a diameter in the order of 500 µm.
MATERIALS DEVELOPMENT, KINETIC CHARACTERIZATION AND EVALUATION (WP4)
Regarding materials development and despite the substantial challenges identified, primarily with respect to the detrimental effect of the cyclic reaction scheme on mechanical stability of pure CaO particles, the progress achieved in the framework of relevant project activities (i.e. work-package 4) was notable.
The kinetics of the relevant reactions have been measured and tested at particle level at the conditions expected in the process. An intrinsic kinetic model was proposed and published. All the material tested in the experimental facilities of the project were characterized and the kinetic model was adapted for the reactor model validation purposes in WP3 and WP6.
Composite particles of CaO/clay and CaO/silicates were developed and successfully evaluated in the course of multi-cyclic hydration/dehydration experiments at lab-scale level. The measured reaction performance of most favorably evaluated compositions was in the range of 40-60% of maximum hydration/dehydration capacity of pure CaO. The most promising compositions exhibited stable cycle-to-cycle reactivity, sufficiently high kinetics and adequate mechanical strength even at the most demanding reaction conditions employed and for up to 200 hydration/dehydration cycles. Such tests were performed via thermo-gravimetric analysis (TGA). At the final stages of the project, one of the most efficient formulations was produced at kg-scale and tested in a batch fluidized bed reactor in the course of 15 hydration/dehydration cycles. The results obtained confirmed previous TGA findings to a large extent and provided further confidence that the approach followed in the duration of the StoRRe project is indeed in the right direction towards the definition of optimized materials to be used in a future scaled-up thermo-chemical storage (TCS) scheme and operated under realistic environment, using the CaO/Ca(OH)2 reaction couple.
EXPERIMENTAL PROOF-OF CONCEPT (WP5 AND WP6)
One of the main objective of the project was to build a continuous bubbling fluidized bed set-up operating in the range of parameters of the kinetics, material and process requirements, the tests campaign would be the basis of the proof-of concept validation. The pilot design and manufacture was completed in 18 month and the starting tests were done in a short time in May and June 2016 with no particular difficulties, which allowed to perform a scarce number but good-quality tests on the pilot before the end of the project.

Figure 1. COCHYSE circuits and reactor after insulation
Prior to the delayed start of the continuous set-up, 78 tests were performed on the batch facility Castorre Chaud, about which 30 are completely reliable and were used for the model validation. A minor number of tests have been done twice and showed a good repeatability. The parameters were the particle size (200 to 800µm), the batch mass (2 to 3.5 kg), the fluidising velocity (0.2 to 0.7 m/s), the bed temperature (400 to 500°C) and the steam ratio in the fluidising gas (0 to 0.8). The tests are performed in transient condition which makes the facility rather difficult to operate and the tests sometimes difficult to interpret.
The continuous bubbling fluidized bed facility aims at studying the chemical reactor of lime hydration and dehydration in steady-state conditions and for a range of parameters larger than the batch facility. In particular, the dehydration can be done under pure steam, and particles as big as 1.5 mm can be fluidized. The facility is designed for a reaction power of 5 kW (the total electric power on the reactor is 20 kW), corresponding to a maximal flowrate of solid of 20 kg/h, it can be operated during 3 hours at the maximal flowrate.
The test campaign on the COCHYSE facility, even if limited because of the late commissioning of the set-up, was successful and proved the concept at pilot scale, the COCHYSE facility is perfectly adapted to the tests requirements. Moreover, the similarity of behaviour was assessed thanks to the reactor models between the batch and the continuous reactor. These first results are very encouraging even if a long way has still to be done before the heat-exchanger reactor concept is demonstrated at industrial scale.
TECHNO-ECONOMIC ASSESSMENT (WP7)
Finally, the best integration of the storage unit in the CSP plant and the development opportunity at the pre-industrial scale has been studied. The thermochemical storage has been compared to other existing storage technologies, particularly in terms of operating and investment costs. It has been concluded from this analysis that the process may have a competitive edge if the large uncertainties that still remain on the operation and cost of some of its key components (reactor-heat exchanger for hydration, reactor-receiver for dehydration) are solved by further R&D work.

Project Results:
PROCESS MODELLING (WP3, CEA)
A simplified scheme of the storage process was modelled in task 3.1 using 2 similar methods, EXCEL sheet or DYMOLA software. The 2 models were quasi-static models even if the parameters could change with time because the masses and volumes of the components were not modelled, so the thermal transitory effects could not be represented. This level of modelling was consistent with the level of knowledge of the storage system, basic interrogations concerning heat and mass balances were questioned.
Several configurations, mainly daily or seasonal storage and pure steam or a mixture of air and steam as fluidising gas, were tested. It was considered that only the chemical energy had to be given to the FB during dehydration, and that only the chemical energy of hydration was delivered to the solar plant, no sensible heat energy was taken into account in the reactor.

Figure 2. Simplified process scheme including a fluidised bed reactor during dehydration (a) and hydration (b) steps with steam as carrier gas.
The modeling of the storage process highlighted some disadvantages of the system. One main issue was the production of steam during hydration, and the energy of steam condensation, that was not valorized during dehydration. The energy to heat up the solid flow was also high, and the compression power for the fluidising gas was not negligible. These points led to a global energy yield of the storage process that was under 50% in all cases, and even below 40% for some seasonal storage. These levels of energy yield were not high enough for a solar plant, and some ideas had to be looked for to better integrate the storage process in the solar plant. A central idea was to valorize the low pressure steam produced during the dehydration somewhere in the solar plant, and probably in the power block sub-system.

Figure 3. Storage process energy efficiency for daily (blue) and seasonal storage (red) in the case where the steam produced by the dehydration is valorised (left) or not (right) in the process
These calculations on the storage model also indicated that it was impossible to operate the dehydration reaction under a very low partial pressure of steam because the dilution of the steam produced by the reaction was too costly, on one hand for the compression energy expenses, and on the other hand for the size of the fluidised bed (large cross-sections required to accommodate the very large gas flows required to maintain low partial pressure of water vapour in the reactors). The reference storage process for Storre was thus a process where both hydration and dehydration were performed under pure steam. This choice gave a simplified process, with fewer heat-exchangers and no need to separate inert gases from steam. For such a process, and considering a residence time of the powder in the reactor of 500 s as a best compromise, the hydration reference temperature given by the intrinsic kinetics (WP4) and reactor model (WP3) is 470°C and the reference dehydration temperature is 540°C. The gap between these 2 temperatures could be reduced by operating hydration under pressure but it makes of course the process more complicated, especially for the reactor feeding in solids and this option was out of the scope of the Storre project.
For our references temperatures and considering only the chemical heat, the temperature of the solar heat transfer fluid (HTF) must be higher than 550°C during dehydration, and will be heated up to a maximum of 460°C during hydration. On one hand it makes the process integration rather complex because the level of the HTF varies between the charge and discharge modes, and on the other hand, the hydration cannot obviously produce steam at a temperature as high as 550°C toward the power block, which was the target.
Concerning the energy density of storage per cubic meter, taking into account only the chemical reaction, it is theoretically very high. With a reaction enthalpy of 104 kJ/mol and a CaO stone density of 3400 kg/m3, the density of storage is as high as 1754 kWh/m3 for solid rocks of calcium oxide. However, lime particles have substantial internal porosity (which is required to accommodate Ca(OH)2 during charge) and needs to be in powder form for a fluidised bed. As a result, a bulk density around 1000 kg/m3 is more realistic which decreases the volume density of storage to 516 kWh/m3. Moreover some inert materiel can be present in the powder and the reaction can be incomplete. For instance, with 30% of inert material and a conversion yield of 60%-which are consistent values considering the composition of the composites manufactured in WP4 and the conversion yield in TGA or Castorre Chaud BFB obtained in WP4- the volume density of storage is only of 217 kWh/m3. This is below the initial target of the Storre project, i.e. 300-500 kWh/m3.
All these considerations led to modify substantially the preliminary storage process design and integration. First of all, the leading idea of having a single heat-exchanger reactor for the 2 reactions of hydration and dehydration was abandoned. As the dehydration had to be done at high temperature, above 550°C, it could be achieved easily only directly at the solar receiver. Solid particles receivers and solid HTF are widely studied but to store sensible heat at high temperature only. Here the idea was to store heat both in sensible and chemical mode. The hydration reactor on its side could directly produce high pressure steam provided that sensible heat again is used to superheat the steam. A side advantage of storing under sensible and chemical mode was also to increase the volume energy density of storage. This new approach is valid for daily (or a few days) storage only as the sensible heat of solid is essential in this process proposal.
REACTOR MODELLING (WP3, CSIC)
Reactor modelling activities in the Storre project were designed as a support for process simulation activities in WP7 and to interpret experimental results from WP6.
In general, reactor models for fluidised bed reactors represent an integration of at least a fluid-dynamic and a kinetic sub-model. The kinetic information to the reactor models developed in Storre come from kinetic sub-models at particle level developed and fitted to experimental results from WP4. Therefore, we only discuss here on the fluid-dynamic model adopted for the different version of the models. In the latest versions this has been the Kunii-Levenspield model (KL model), which is a model that assumes perfect mixing of solids and two gas phases in the reactor (a bubble phase free of solids and a dense emulsion phase of solids maintained at minimum fluidization conditions). A certain crossflow of gas to account for the exchange of reacting gas species between the phases can be defined in the model.
The KL model has been adapted in the Storre project for the hydration and dehydration reactions by introducing a kinetic sub-models suitable for both reactions. In some early applications of the reactor model for process design of early hydration/dehydration concepts, a plug flow model for the gas phase (equivalent to a large gas exchange coefficient between phases) has been adopted. However, when using the model to interpret experimental results from Castorre Chaud and Cochyse test facilities, the use of the cross flow factor as adjustable parameter has been the key to interpret the experimental observations. A single value of 1.5 for this dimensionless parameter is able to make predictions that are consistent with the experimental results as long as the specific kinetic parameters at particle level are incorporated to the model from independently measurements of rate of reactions in a thermogravimetric apparatus. Furthermore, it has been observed that the sensitivity of the model predictions to the value of the crossflow factor is small when fitting experimental results of hydration (in Cochyse or Castorre Chaud). This is because the modest rate of hydration of the CaO material used in these test (CaO from the calcination of CaCO3 at T>1000ºC). However, the predictions are more sensitive to the crossflow factor in dehydration experiments, where the kinetics at particle level are relatively faster and the resistances to the progress of reaction from the bubble-emulsion exchange coefficient become more important. Therefore, for materials with higher intrinsic reactivity (as those synthetized in WP4) the performance of the reactor (in terms of solid and gas conversions as a function of inputs, dimensions and operating conditions) is likely to be determined/controlled largely by fluid-dynamic aspects related to the bubble-emulsion interaction, that may require more accurate modelling when considering the large changes in the gas phase due to the disappearance (during hydration) or appearance (during dehydration) of a large fraction of the fluidising gas.
Regarding foregrounds, there are still two main reactor concepts to be considered for more detailed modelling to aid in the scale up towards TRL6-7 and beyond (both of them envisaged to treat large flow of gases as required to keep reactor intensities in the range of the MWth/m2 of reactor cross-sections):
− Bubbling fluidised beds of large particles (mm range). This has been the main demonstration activity at pilot level (TRL3 and 4) in the Storre project. It requires mechanically stable particles (attrition resistant) containing active CaO (not yet fully available). Large particle sizes are preferable to maximize fluidisation velocities with minimum entrainment. Dense beds (relatively swallow) lead to large cross sections and reactor lengths, as necessary to accommodate the required volume for heat transfer surface, while maintaining a moderate ΔP in the bed. Investigation of fluid-dynamics aspects to enhance axial mixing of solids and minimise lateral mixing in the resulting long bed (thereby allowing for more efficient countercurrent heat transfer flows) is an important issue for future scaling up.
− Circulating fluidised bed of fine particles (< 300 micron). This is an option conceptually investigated in Storre, showing substantial potential advantages for scaling up at large scale. Several elements of commercial CFB boilers (available at the range of 1000 MWt using CaO similar materials) are well known: cyclones, standpipes external fluidised bed heat exchanger. The use of high gas velocities (3-7 m/s) allow for compact reactor designs (smaller cross section), recycle/circulation of solids allows for sufficient solid residence times and external heat transfer, relatively low ΔP are possible because the reduced solid inventories, etc. In contrast, there are substantial challenges for exploitation of these advantages: enhanced attrition because of higher velocities, the lack of suitable material and/or technology to handle extremely fine particles generated during repetitive hydration-dehydration cycles, higher complexity in the full reactor set up and scalability (although much experience could be adapted from similar technologies using CaO in CFBs).
HEAT-EXCHANGER REACTOR PRE-DESIGN (WP3, CEA)
For the hydration reactor, the general idea was to find a very simple design in order to minimize the difficulties with the solid handling and hydrodynamics. Therefore a concept with a counterflow heat-exchanger of a single-pass tube bundle-calender kind was proposed. The objective was to use both the chemical and the sensible heat of the solid.

Figure 4. Scheme of a simple counterflow heat-exchanger reactor with a single tube bundle
Some assumptions on the solid flow are done in the model, the solid emulsion is considered as perfectly mixed in a section and has a plug flow circulation along the bed.
The commercial plant specification is to have a constant power block of 85 MWe whatever the period of the day. For the hydration heat-exchanger reactor, the dimensioning condition is always the discharge mode. The power delivered to the HP steam is higher (213 MWth) and the reactor must provide moreover the LP steam for the hydration reaction. This LP steam production is indeed a main concern, and 3 configurations are tested, in the 1st one the LP reaction steam is directly produced in contact of the solid bed, in the 2nd one it is produced in a second BFB system parallel to the main one (see Figure 5), and in the 3rd one it is produced in series with the main HP system.

Figure 5. Reactor scheme - the reacting LP steam is produced in a parallel BFB
The 3 solutions are similar in terms of bed volume and fluidising costs, the second solution is probably the better one, even if 2 BFB are needed, because the main HP bed system is operated at constant power (202 to 213 MWth) in charge and discharge mode.
The design is done for the following parameters:
− The tubes arrangement is staggered
− The horizontal pitch of tubes is 2*Dext,hyd, the more compact value that can be chosen.
− The vertical pitch of tubes is 2*Dext,hyd, the more compact value that can be chosen.
− The distance between the distributor and the first row of tubes is 5*Dext,hyd
− ½ inches tubes Schedule 80 (21.3 and 15.76 mm external and internal diameters)
The 1D model shows that the hydration occurs at variable temperature, as highlighted on Figure 6, and that the hydration conversion is completed at the end of the reactor. Even with the compact values taken for the bundle, the bed volume is rather large and the solid residence time (~1000 s) is more to what is necessary for kinetics needs only because the global heat transfer coefficient is moderate in the range 250-400 W/m2/°C on the bed length, the HP steam tube bundle dimension will give the bed dimension more than the reaction kinetics, it explains also why the hydration is complete at the end of the bed. The main advantage is that the bed design is not very much impacted by a possible slow-down of the hydration kinetics, the size is the same at full speed or for a kinetics slowed by a factor 5.

Figure 6. Solid and Steam profiles of temperatures in the HP steam power x-axis (1D model)
The hydration reactor is fluidised under pure steam flowing in a closed loop, a compressor must compensate the pressure drops of the circuit. The compression power must in any case be much lower than the CSP block power. This electric power is proportional to the volume of the bed and to the fluidising velocity (or particle size). The superficial velocity in the hydration bed must be above the minimal fluidisation velocity Umf everywhere, at the bottom as at the top. As steam is consumed in the bed, the velocity at the outlet is always lower than at the inlet. In the model, we assume that the steam velocity is regulated at the outlet with a constant ratio with Umf, and that enough reaction steam is fed at the bottom in each mesh of the model for the reaction. It allows to draw the evolution of the reaction flowrate and fluidising flowrate for each mesh of the model, the cumulated flowrates and the corresponding superficial velocities along the bed length. For particles size of 500µm, the 2 cumulated flowrates are quite similar, but the repartition along the bed is quite different. At the maximum at the length +60 m, the inlet velocity is only half of the transport velocity.

Figure 7. Reaction and fluidising flowrates per mesh, cumulated along the bed length, and superficial velocities for particles of 500 µm and Factor,kinetic,hyd=1
Feeding the reaction steam at the bottom of the bed is indeed a factor that is determinant in the reactor design, even with a bed volume that is already bigger than needed for the reaction kinetics itself. The base section must be as large as possible and the particle must have a minimal size (> 300 µm) to avoid transportation. The velocity must be controlled at the outlet at several positions because the steam flowrate consumed by the reaction varies a lot along the bed length. It is advantageous to have a hydration reaction that is not too fast under pure steam because it mitigates the velocity pic.
For particles of 500µm, the thermal compression power varies from ~14 MWth if the steam is maintained hot at the bed outlet mixing temperature to ~8MWth is the steam is cooled down to 150°C, it is then interesting to add a cooling/heating heat-exchanger on the fluidising gas circuit in order to compress steam at low temperature. In that case, the compression power is less than 4% of the turbine power, which is acceptable and validates the choice of a fluidised bed concept that is sometimes criticized for the excessive power losses due to fluidisation.
MATERIALS DEVELOPMENT, KINETIC CHARACTERIZATION AND EVALUATION (WP4, APTL + CSIC)
At the initial stages of the project, it was made evident that the most prominent challenge to be tackled with respect to materials development activities was the severe fragmentation of CaO/Ca(OH)2 particles upon cyclic hydration/dehydration. This was mainly attributed to the substantial swelling of CaO grains during transformation to Ca(OH)2. Such a phenomenon could be detrimental to the operation of fluidized bed reactors, proposed by StoRRe for the particular TCS scheme, as it would lead to the generation of substantial amounts of very fine particles thereby making fluidization practically impossible. Thus, the use of pure CaO was disqualified from further stable materials development studies although they were still considered adequate for kinetic studies and single cycle use during batch and continuous pilot experiments, due to their commercial availability. In this respect, a methodology was developed and applied to determine kinetic parameters from TGA measurements, under realistic temperatures and partial pressures of steam for the Storre process. The resulting reaction model of CaO or Ca(OH)2 grains at particle level was applied to derive intrinsic kinetic parameters used as a benchmark of other synthetic materials. Also, it was a useful sub-model adaptable to fit reactivity measurements from pure compounds to composites of CaO grains or highly sintered commercial CaO batches uses in the pilot test.
The main focus to develop synthetic CaO functional materials for TCS was on the identification of abundant, non-toxic, cost-effective and refractory (ceramic) materials to be used as additives/binders and ensure mechanical robustness of CaO-based formulations in the course of as many as possible hydration/dehydration cycles.
Initial attempts focused on the development of CaO/Al2O3 composites via liquid-synthesis routes. Measurable, albeit not adequate, improvement with respect to mechanical stability cf. pure CaO was identified. This was mainly attributed to the formation of mixed calcium-aluminum oxides during high temperature calcination. Naturally, the addition of Al2O3 caused a decrease in hydration capacity versus pure CaO and it was found that such a decrease was actually over-proportional to the Al2O3 amount added. Preliminary evaluation studies indicated the formation of mixed hydrated species with appreciable energy density on the basis of relevant DSC measurements. This was initially considered as an important finding because the particular species could be formed/decomposed at substantially lower temperatures cf. CaO/Ca(OH)2, thereby providing more flexibility in terms of operating temperatures of the cyclic hydration/dehydration scheme. Due to the fact that the formation of mixed hydrated species could not be identified under more realistic reaction conditions and also considering the insufficient mechanical stability of such CaO/Al2O3 composites, the particular compositions were eventually disqualified.
In an attempt to focus on the development of pellets with high crushing strength, the use of sodium silicate as an additive to promote a matrix of hard calcium silicates in the pellet was investigated. The experimental screening of optimum conditions to maximize crushing strength and CaO reactivity on hydration at the maximum level of hydration-dehydration cycles resulted in a patent application.
The next steps involved evaluation of several additives, in-principle meeting the above mentioned requirements. Prominent examples included (nano)silica, fly ash, mullite, high temperature cements and various silicates and clays. The most promising results were obtained when using specific grades of silicates or clays, while it was also identified that the preparation protocols of such CaO-based composite formulations were of crucial importance. Parametric studies included the nature of CaO-precursor and synthesis route, the calcination temperature, heating/cooling rates and atmosphere employed, the particle size distribution of raw powders, the wt% of additive/binder employed etc. As expected, the main structural stabilization mechanism was due to the formation (during calcination) of a mechanically stable and relatively chemically inert binary/ternary mixed oxide matrix via the solid state reaction of CaO and the additive used. Free CaO was able to participate in the cyclic hydration/dehydration steps without notable inhibitory effects regarding reactivity and reaction kinetics. Initially, materials were prepared and evaluated in the form of particles with sizes in the range of 1-2.5 mm. The main key performance indicator values of the optimized CaO/silicate and CaO/clay compositions, as well as other important findings identified in the course of the development/characterization/evaluation path employed are summarized as follows:
− Measured hydration/dehydration capacity values of qualified composites was in the range of 40-60% of maximum (theoretical) hydration capacity of pure CaO. CaO/silicate composite showed somewhat higher hydration/dehydration capacities than the best CaO/clay compositions.
− 20-200 cycle hydration/dehydration TGA tests resulted in crushing strength (CS) values of best performing composite used particles in the range of 5 - >30 N. The respective CS values for fresh formulations were between 10–30 N. In general, best performing CaO/clay compositions showed measurably better mechanical stability behavior. For direct comparison purposes, it should be noted that even most mechanically stable (i.e. via very high temperature calcination) pure CaO grades used as benchmark in the project showed CS values of virtually 0 N in the course of just 10-20 cycles. Naturally, the CS values measured were directly proportional to the composite particle size (in general particle sizes in the range of 800-2500 μm were studied).
− Attrition tests in a dedicated setup provided further proof that the CaO/clay composite particles are of higher mechanical stability cf. CaO/silicate ones. For the former case, 5 h attrition experiments at ambient temperature resulted to non-measurable / detectable fines generation.
− The storage conditions of produced particles were found to be of prime importance with respect to composite particles stability. The materials tend to be slowly hydrated (primarily) and carbonated (secondarily) by atmospheric moisture CO2 respectively, with the former proving to be detrimental to structural stability and the latter rendering part of CaO inert (as CaCO3 decomposition is well above the effective operating temperature of hydration/dehydration scheme). This phenomenon was much more profound for the case of CaO/clay formulations and storage at low vacuum conditions was found to be not very effective. Most of the times, even the most stable such compositions collapsed completely in the course of 20 cycles for cases that cyclic hydration/dehydration was not performed just after calcination. Potential solutions (in-project validated) included: a) transportation/storage of the material in non-calcined form and calcination on-site to minimize the period between calcination and use and b) transportation/storage of the material in hydrated form under CO2-free conditions.

Figure 8. Photograph of CaO/clay composite particles (left) and long-term (200 cycles) hydration capacity of the same sample, as measured by TGA.
One of the most promising formulations (i.e. CaO/clay) was produced at a quantity of approximately 1.8 kg and after calcination was tested in a small-scale batch fluidized bed reactor. The size distribution of the (nearly-spherical shaped) particles used was in the range of 500-2000 μm (global mean particle size was 1 mm). The material was subjected to 15 hydration/dehydration cycles. Based on the reactivity measurements and post-examination/physico-chemical characterization of used particles, the following conclusions were extracted:
− The formulation was sufficiently reactive, in-line with previous TGA findings. Measurable cycle-to-cycle deactivation was identified but, as confirmed by relevant post-analysis, it was due to gradual CaCO3 formation because the fluidized bed setup had no particular measures for removing ambient CO2 from the system.
− Measured CS values of particles in the bulk of the bed were in very good agreement with previous relevant findings on used particles subjected to 20 hydration/dehydration cycles (5 - >30 N depending on particle size). Post examination and sieving of used particles revealed that the amount of generated fines (I.e. size < 75 μm) was below 0.5% of total material mass used in the reactor.
In the framework of WP4 activities, a kinetic model was extracted on the basis of intrinsic activity measurements of CaO formulations in a TGA setup. The model was further refined and validated with experimental data at the batch fluidized bed reactor. The agreement between the model and the experimental data was very good.
The above summarized promising results indicate significant improvement with respect to suitable CaO-based formulations development, within the duration of StoRRe project. Further optimization studies by using WP4 findings as starting point have already been defined to be implemented in the framework of potential future follow-up activities.
EXPERIMENTAL PROOF-OF CONCEPT (WP5 AND WP6, CEA)
One of the main objective of the project was to build a continuous fluidized bed set-up operating in the range of parameters given by the kinetics, material and process requirements, the tests campaign would be the basis of the proof-of concept validation. The detailed design and manufacture of the continuous pilot (WP5) was unfortunately delayed by the termination of the 2 industrial partners (SHAP and AREVA Renouvelable) and some technical difficulties. The pilot design and manufacture was completed in 18 month and the starting tests were done in a short time in May and June 2016 with no particular difficulties, which allowed to perform a scarce number but good-quality tests on the pilot before the end of the project.

Figure 9. COCHYSE circuits and reactor after insulation
Prior to the delayed start of the continuous set-up, 78 tests were performed on the batch facility Castorre Chaud, about which 30 are completely reliable and were used for the model validation. A minor number of tests have been done twice and showed a good repeatability. The parameters were the particle size (200 to 800µm), the batch mass (2 to 3.5 kg), the fluidising velocity (0.2 to 0.7 m/s), the bed temperature (400 to 500°C) and the steam ratio in the fluidising gas (0 to 0.8). The tests were performed in transient condition which makes the facility rather difficult to operate and the tests sometimes difficult to interpret.
The continuous bubbling fluidized bed facility aims at studying the chemical reactor of lime hydration and dehydration in steady-state conditions and for a range of parameters larger than the batch facility. In particular, the dehydration can be done under pure steam, and particles as big as 1.5 mm can be fluidized. The facility is designed for a reaction power of 5 kW, corresponding to a maximal flowrate of solid of 20 kg/h, it can be operated during 3 hours at the maximal flowrate.

Figure 10. Schematic view of the continuous experimental setup.

The reactor is similar to the batch BFB reactor and is a cylinder 108 mm of diameter and 780 mm of height, with no disengagement zone, it is made of Inconel 600 alloy to sustain temperatures as high as 800°C on the walls and is designed for a maximal pressure of 1.5 bars. It is heated by 2 main electric heaters of 6.5 kW each (Heaters 2 and 3) and a smaller one of 2.5 kW (Heater 4) and several other smaller heaters that compensate heat losses and preheat the inlet solid, the total heating power on the reactor is 20 kW. The reactor bed is cooled down by a submerged heat-exchanger designed to remove 4 kW in the worse conditions, the heat transfer fluid (max 60 Nm3/h) is cold or warm air. The fluidizing gas is either pure air (max 40 Nm3/h) or pure steam (max 32 kg/h) or a mixture of these two gases.
The solid powder is continuously and steadily fed during several hours by a feeding system including a hopper of 80 liters, a cold regulating feeding screw, a rotary valve and a warm transport screw connected to the lower part of the reactor. A tight rotary valve prevents the hot steam and air to flow from the reactor to the hopper. In the hopper, a screw prevents the arching risks. The mixture of solid and gas is extracted from the reactor by a diluted pneumatic line cooled by natural convection and connected to a high temperature filter (max 200°C) where the gas is separated from the solid. To have a constant transport velocity close to 20 m/s, hot air (max 90 Nm3/h) is added to the mixture of gas and solid at the outlet of the reactor, this addition of air helps to cool down the gas-solid flow before the filter and to avoid condensation when the reactor is operated under pure steam. The solid is continuously removed from the filter volume by a system including a rotary valve and a hopper. The tight rotary valve prevents the hot air and steam to flow from the filter to the hopper. The technological solutions to move the solid might not be the best one at industrial scale, especially the diluted pneumatic transport that drives the solid from the reactor to the filter is indeed not the best economic choice. Many other solutions could be preferred at large scale, such as mechanical transport or dense phase transport, which is not feasible for a reactor with a volume of 7 litres and a solid flowrate of 20 kg/h. The pilot technology is therefore not a pre-design of any industrial solution and was designed for the pilot scale only.
After the filter, 2 hygrometers measure the gas moisture and allow, by difference with the inlet steam, to calculate the instantaneous reaction conversion yield. As the COCHYSE facility is operated in steady state conditions, the evolution of the reaction can be predicted also by a thermal balance, generally there is a good agreement (+-0.05%) between the 2 methods. The solid flowrate is measured by the rotation speed of the regulating screw after calibration tests on one hand and by the weight sensors set on each hopper on the other hand. Three thermocouples type K at 4 levels measure the bed temperature, 2 thermocouples type K measure the external wall temperature near each heater, the pressure is measured before the gas distributor, at 4 levels in the bed and after the filter. The electric power of all the heaters is measured. The cooling power of the heat-exchanger is calculated from the air flowrate and the inlet and outlet air temperatures. A sampling line allows to sample solid from the reactor during a test and to control the conversion yield by weighing the sample before and after full dehydration in a laboratory oven.
The lime provided for the continuous tests is a commercial grade, 95%w CaO, supplied by CARMEUSE, obtained by the calcination of CaCO3 at 1000°C and sieved to separate the fine and big particles, the particle size range is 200-800 µm. The chemical behavior of this lime was tested by CSIC and is similar to the lime used during the batch tests. 1000 kg of calcium oxide CaO were delivered in 5 tight barrels of 200 kg each.
10 tests have been done in total, 5 are relevant (Hydrations #4, #6, #8 and dehydration #3 and #7) and were compared to the reactor model, the tests marked ND (Not Done) should be done in the future after the end of the project in view of a publication:
Table 1. Relevant hydration and dehydration tests
Hydration
Solid Flowrate (kg/h) 10 10 10 10 18
Fluidizing velocity (m/s) 0.4 0.6 0.6 0.6 0.6
Steam molar fraction (-) 1 0.5 0.75 1 1
Temperature (°C) 400 ND #2*
#8*** #6*** #4*** ND
450 ND
475 ND
Dehydration
Solid Flowrate (kg/h) 12 12 12 12 20
Fluidizing velocity (m/s) 0.4 0.6 0.6 0.6 0.6
Steam molar fraction (-) 0 0 0.5 1 0
Temperature (°C) 550 ND #3** #7** ND #5** (fresh lime)
#9** (fresh lime)
530 ND

The evolution of the regulated parameters, steam flowrate, solid flowrate, bed temperature and bed heating and cooling powers are given on Figure 11 for a typical hydration (#4) and dehydration reaction (#3). Contrary to the batch reactor, the parameters are constant during the tests apart from the very beginning and end. In particular, thanks to the cooling heat-exchanger, the hydration reaction is constant all along the test which could not be achieved on the batch reactor:

Figure 11. Evolution with time of the inlet (blue) and outlet (green and cyan) steam flowrates, solid flowrates (red) , bed temperatures (mean value on 4 levels in the bed), electric power (red for the bed and orange for the screw) and cooling power (green) : Hydration #4 (left) and Dehydration #3 (right)
The test campaign on the COCHYSE facility, even if limited because of the late commissioning of the set-up, was successful and proved the concept at pilot scale, the COCHYSE facility is perfectly adapted to the tests requirements. Moreover, the similarity of behaviour was assessed thanks to the reactor models between the batch and the continuous reactor. These first results are very encouraging even if a long way has still to be done before the heat-exchanger reactor concept is demonstrated at industrial scale.
TECHNO-ECONOMIC ASSESSMENT (WP7)
Task 7 global objective is to evaluate the techno-economic feasibility of the system developed integrated in a commercial scale thermosolar power generation facility. It is structured in three parts: i) definition of the power generation philosophy and simulation parameters; ii) system integration; and iii) techno-economic evaluation and feasibility analysis.
The first part of the task (WP7.1) is to define the simulation details and the general lines of the operation philosophy of the plant in order to generate the annual performance model for Task 7.3. The power and capacity of the storage system aim to be simulated are also established. Once the final plant configuration is closed (in WP7.2) the simulation and evaluation procedure of the technology proposed will be very similar to the procedures applied to commercial plants in order to optimize their configuration and specifications. The procedure consists in the following main steps:
− Power block simulation
This process is done with Aspen Plus assuming theoretical efficiencies for the different blocks, theoretical properties of the working fluid and balancing the system trying to achieve the maximum global efficiency.
− Steam generator/Reactor simulation
Those calculations are performed with Excel and EES performing several iterations between the preliminary sizing of the heat exchanger and adjusting the temperatures of the different streams in order to get a proper design matching with the kinetics of the material and heat transfer area needed.
− Performance model simulation
The performance model is carried out in Matlab and several iterations are needed to obtain the optimum plant configuration in terms of energy generation cost along the lifetime of the plant.
From the three steps described below an optimum plant configuration is specified calculating the desired parameters to be able to mark the different plant specifications.
There are also defined the main operation modes to reproduce as much as possible the operation strategies of the Concentrated Solar Power (CSP) commercial plants currently in operation in order to obtain results as close to the reality as possible.
Finally the location, the power of the plant object to study and the thermal capacity of the storage system are fixed depending of the available Direct Normal Irradiation (DNI) and the different profiles of electricity demand. It is concluded to study two different thermal capacities in order to see the effect of increasing the storage system capacity. Options object to study are:
− Power 85MWe; capacity 6h of storage (510MWhe)
− Power 85MWe; capacity 13h of storage (1105MWhe)
It is important to integrate the system within a plant configuration in the most efficient manner as possible. Afterwards that would allow to compare fairly the results with similar technologies already present to the market.
Direct Steam Generation (DSG) based in tower CSP technology is selected for the integration of the system proposed due to the temperatures of the reaction match well with those cycles. Once decided that a preliminary plant diagram is proposed, later a temperature level and pinch analysis is performed on the main heat exchanger. Next step is to study several Bubbling Fluidized Bed (BFB) heat exchanger configurations depending on the way to integrate the steam power cycle with the TCES material streams in the mentioned heat exchanger. Finally, operation rates and flowrates of this BFB are calculated for the different operation modes designed for normal operation.
After being analyzed the BFB, several other plant configurations (plant diagrams) are also analyzed trying to maximize the energy usage and efficiency; all the needed components are defined conceptually to see all the integration possibilities of each one. After being analyzed preliminary performances and energy integration of all of them one final configuration is selected. This plant configuration (plant diagram) chosen is developed integrating all the streams and defining the components interconnection with preliminary energy balances for the main operation modes.
The main operation modes of the plant configuration are defined in order to be able to implement them in the annual performance model. Global efficiencies, flowrates and load ratios of the plant and the power production are calculated. This is developed integrating the TCES material energy balances and the power block one with different software. There are calculated the efficiencies, operating rates of the power block, operating rates of the heat exchangers and the different flowrates of the streams of the plant. All is calculated at different situations depending on DNI available and status of storage system which are the basis for the annual performance model calculations.
With all the data ready to run the annual performance model the next step is to determine the potential of the proposed technology to be introduced into the market. This is done analyzing the performance of a commercial scaled-up plant based in the technology proposed and comparing it with the benchmark of the Concentrating Solar Power (CSP) alternatives existent to the market. Furthermore safety, carbon footprint issues and others are also considered.
The first step is to up-scale the plant to the commercial size established. Seven different sizes of solar field combined with two different storage capacities are evaluated economically. Those constitute seven different plant configuration options (combinations of solar field area and storage system thermal capacity) in terms of thermal capacity and solar field area. At the end, the optimum configuration for each storage system thermal capacity is chosen. This represents to scale the plant to 85MWe with two different capacities of storage: one for 6h and the other for 13h.
Next, the performance of each plant is calculated. A sensitivity analysis vs. the cost of the receiver is performed to check the importance of the variability of the most volatile component of the plant in terms of cost. As commented before the plant configuration is optimized in terms of solar multiple and thermal capacity in order to get the most competitive costs of LCOE.
Once the LCOE and the cost breakdown of the solution are calculated a cost comparison with other similar technologies is done in terms of LCOE and storage system costs. Different aspects related with carbon footprint and other environmental factors are also considered. Finally a risk analysis of the solution is presented detailing the points of weakness and the proposed countermeasures to avoid them.

Potential Impact:
Regarding reactor modeling and experimental activities (WP3, WP5 and WP6), the only impacts and dissemination and exploitation activities that can be discussed are of scientific nature. The work carried out in this respect within the Storre project has demonstrated the suitability of the KL model for bubbling fluidised bed reactors. This finding has been disseminated in the following scientific publications:
− Criado, Y. A.; Alonso, M.; Abanades, J. C., Anxionnaz-Minvielle, Z. Conceptual process design of a CaO/Ca(OH)2 thermochemical energy storage system using fluidized bed reactors. Appl. Therm. Eng. 2014, 73, 1089.
− Criado, Y. A.; Huille, A.; Rougé, S.; Abanades, J. C. Experimental investigation and model validation of the CaO/Ca(OH)2 fluidized bed reactors for thermochemical energy storage applications. Chem. Eng. J. 2016, Submission date: 15 June 2016.
− Rougé, S.; Criado, Y.A.; Huille, A.; Abanades, J.C. Proof of concept of the CaO/Ca(OH)2 reaction in a continuous heat- exchanger BFB reactor for thermochemical heat storage in CSP plants, SolarPaces 2016. Accepted for oral and AIP Proceedings:
A fourth scientific publication between CEA/CSIC is in preparation to report on the fitting of the continuous version of the KL model to experimental results in the continuous pilot Cochyse.
Regarding StoRRe achievements with respect to the development of sufficiently reactive, high mechanical strength and cost-effective CaO-based formulations (WP4), the main project findings have clearly contributed towards mitigating one of the main barriers towards future pre-commercialization of the technology proposed by StoRRe; namely the lack of sufficiently mechanically stable materials to be employed in multi-cyclic hydration/dehydration systems. The main developmental paths generated by StoRRe are directly applicable to several other environmental-friendly application fields requiring the use of highly robust particles in moving concepts (fluidized beds, falling particles etc.). Examples include materials for CaO/CaCO3 fluidized beds (either for CO2 capture or TCS applications) and falling particles as heat transfer fluids in very high temperature future solar receivers. The partners responsible for such activities (i.e. APTL/CERTH and INCAR-CSIC) have gained substantial know-how and at the end of the project it can be considered that they have gained a significant competitive advantage in the field.
Dissemination and exploitation outputs are as follows:
− Criado, Y.A. Alonso, M., Abanades, J.C. 2014. Kinetics of the CaO/Ca(OH)2 hydration/dehydration reaction for thermochemical energy storage applications. Industrial & Engineering Chemistry Research 53, 12594−12601.
− Sakellariou, K.G. Karagiannakis, G., Criado, Y.A. Konstandopoulos, A.G. 2015. Calcium oxide based materials for thermochemical heat storage in concentrated solar power plants. Solar Energy 122, 215-230
− Abanades, J.C. Criado, Y.A. Alonso M., 2015. Method for preparing a thermochemical energy storage material. EP15382078.2.
− Criado, Y.A. Alonso, M., Abanades, J.C. 2015. Composite material for thermochemical energy storage using CaO/Ca(OH)2. Industrial & Engineering Chemistry Research 54, 9314-9327.
− Criado, Y.A. Alonso, M., Abanades, J.C. 2016. Enhancement of CaO/Ca(OH)2 composite material for thermochemical energy storage. Solar Energy 135, 800–809.
− K.G. Sakellariou, N.I. Tsongidis, G. Karagiannakis, A.G. Konstandopoulos, 2016. Composite CaO-based structured bodies for thermochemical heat storage with the CaO/Ca(OH)2 reaction scheme. Part 1: Shortlisting of suitable materials and reparation methods, submitted to Solar Energy journal, Article under review
− K.G. Sakellariou, N.I. Tsongidis, Y.A. Criado, G. Karagiannakis, A.G. Konstandopoulos, 2016. Composite CaO-based structured bodies for thermochemical heat storage with the CaO/Ca(OH)2 reaction scheme. Part 2: Multi-cyclic evaluation, In preparation

List of Websites:
http://storre-project.eu/

Final Report Summary - STORRE (High temperature thermal energy Storage by Reversible thermochemical Reaction)

Share this page

Download